Cutting Edge: TLR4 Deficiency Confers Susceptibility to Lethal Oxidant Lung Injury¹

Inhalation of high levels of oxygen (hyperoxia) is a necessary and life-saving component in the treatment of critically ill patients. However, prolonged hyperoxia leads to excessive oxidant stress via the accumulation of reactive oxygen species (ROS)³ that initiate epithelial and endothelial cell death, increased pulmonary capillary permeability, inflammation, lung destruction, and ultimately death (1). ROS have been linked to TLR4 activation and signaling (2). Functional TLR4 has recently been described in the lung (3), but its role in the lung is poorly understood. The lungs, unlike many other organs, are constantly exposed to both microbial agents as well as ambient oxygen and have therefore evolved an extensive system of innate immune and antioxidant defenses, previously thought to be distinct pathways. Our data suggest that TLR4 may mediate a common pathway for innate immune, antioxidant, and antiapoptotic responses.

Generally, TLR4 deficiency is thought to be protective against models of injury such as endotoxin, ischemia-reperfusion, and ozone (4, 5, 6). To the best of our knowledge, we demonstrate for the first time that TLR4 is essential for survival in vivo and in lung structural cells during lethal hyperoxic exposure. Furthermore, we use exogenous delivery of heme oxygenase (HO)-1 as both an antioxidant and antiapoptotic strategy to successfully rescue TLR4^−/− mice from oxidant-induced death. HO is the rate-limiting enzyme that degrades heme into bilirubin, free iron, and CO (7). Three isoforms of HO exist: HO-2 and -3 are constitutively expressed, whereas HO-1 is the inducible isoform and thought to function as an antioxidant (8). The ability of HO-1 to restore Bcl-2 and phospho-Akt protein expression as well as improve survival in TLR4^−/− mice highlights the role of TLR4 in maintaining antioxidant and antiapoptotic balance in the lung.

TLR4^−/− (B6;129Tlr4^tm1Aki) and the control mice (B6;129F2) have been described previously (9). The TLR4^−/− mice were originally provided by S. Akira (Osaka University, Osaka, Japan) (4). Mice were maintained and bred under specific pathogen-free conditions at the animal facility of Yale University School of Medicine. All of the protocols were reviewed and approved by the Animal Care and Use Committee at Yale University.

Mice were exposed to 100% oxygen (O₂) in a Plexiglas exposure chamber. Naive mice were kept at room air. For survival studies, animals were carefully monitored and the time of death noted. Lung injury was assessed 72 h after the initiation of hyperoxia by performing bronchoalveolar lavage (BAL) cell counts and protein analyses. Lung specimens were also processed for histology, RNA and protein extraction, apoptosis, and immunohistochemistry analyses.

Lung endothelial cells and type II alveolar epithelial cells were isolated from lungs of wild-type (WT) and TLR4^−/− mice with modification of the methods described previously (10). Hyperoxic conditions were achieved as described previously (11).

TUNEL assay was performed, and TUNEL-positive cells were expressed as a percentage to total cells as described previously (11). Flow cytometry was performed on cells using a FACS to detect Annexin V-FITC labeling (BD Pharmingen) as described previously (12).

Protein levels of Bcl-2, Bad, Bax, caspase-3 (Santa Cruz Biotechnology), HO-1 (StressGen Biotechnologies), and phospho-Akt (Cell Signaling Technology) were analyzed by Western blot assays as described previously (13).

Mice were anesthetized with methoxyflurane, and then 5 × 10⁸ PFU of adenoviral HO-1 (Ad-HO-1) (a gift from L. E. Otterbein, Harvard University, Boston, MA) or adenoviral β-galactosidase (Ad-LacZ) (BD Biosciences) were administered intranasally to each mouse in a volume of 50 μl as described previously (12).

Total RNA from lung tissue was extracted as described previously (12). For mouse TLR4, sense: GCTTTCACCTCTGCCTTCAC; antisense: CGAGGCTTTTCCATCCAATA; and for mouse β-actin, sense: GTGGGCCGCTCTAGGCACCAA; antisense: CTCTTTGATGTCACGCACGATTTC. RT-PCR was performed using RT-PCR Master Mix (USB).

Data are expressed as mean ± SE and analyzed by Student’s t test. Survival studies were evaluated using log-rank analysis (14), and statistical analysis was performed using GraphPad Prism 3.0 (GraphPad) software. Significant difference was accepted at p < 0.05.

To investigate whether TLR4 and its endogenous ligands are involved in hyperoxia-induced lung injury, we exposed WT mice to hyperoxia for 72 h, a time point previously determined to be most representative of maximal stress responses in the murine lung (15). We found that hyperoxia increased TLR4 mRNA expression in whole lung lysates (Fig. 1,A). We also confirmed increased TLR4 protein expression by immunohistochemistry in a variety of lung cells (including lung epithelial and endothelial cells) after hyperoxia (data not shown). Hyperoxia also increased expression of the endogenous TLR4 ligands fibronectin and hyaluronan in both WT and TLR4-deficient mice (TLR4^−/−) (data not shown). To determine whether there was a functional role in vivo for TLR4 induction during hyperoxia, we assessed survival rates of WT and TLR4^−/− mice in continuous hyperoxia. Given that TLR4 deficiency has been shown to be protective in other models of noninfectious injury (4, 5, 6), we expected TLR4^−/− mice would be less susceptible to the damaging effects of hyperoxia. To our surprise, TLR4^−/− mice showed significantly increased mortality during hyperoxia compared with WT mice (Fig. 1,B). After 5 days of hyperoxia exposure, 55.6% of the WT mice (n = 18) remained alive, whereas none of the TLR4^−/− mice (n = 18) were alive. The range of survival for TLR4^−/− mice was 2.5–5 days, and the range for WT mice was 3.5–6.5 days. This indicated that a TLR4-dependent signaling pathway was critical for survival during hyperoxia. Consistent with the survival data, TLR4^−/− mice exposed to hyperoxia exhibited significantly greater inflammation, lung permeability, and oxidative DNA damage compared with WT mice (Fig. 1, C–E). Of note, naive TLR4^−/− mice and naive WT mice have similar basal levels of BAL cell counts, protein, and DNA oxidation (data not shown).

Both animal and recent patient studies have established apoptosis, specifically lung endothelial and epithelial apoptosis, as an important part of the pathogenesis of acute lung injury (16, 17). We found that TLR4^−/− mice exhibited significantly greater lung TUNEL staining (Fig. 2,A). Both epithelial and endothelial cells isolated from TLR4^−/− mice also showed significantly more apoptosis than cells from WT mice during hyperoxia (Fig. 2, B and C).

We postulated that a potential mechanism of increased lung apoptosis in TLR4^−/− mice might be differences in anti- and proapoptotic protein expression. Our Western blot results showed that hyperoxia increased expression of Bcl-2 in WT lungs (Fig. 3). However, TLR4^−/− mice were unable to significantly increase Bcl-2 protein in their lungs. The expression of Bad and Bax appeared relatively unchanged during hyperoxia in the WT and TLR4^−/− mice. The Akt pathway is also known to play an important role in modulating apoptosis and was recently linked to TLR4 in response to LPS (18, 19). We found that TLR4^−/− mice were unable to increase phospho-Akt expression in response to hyperoxia, unlike WT mice (Fig. 3). In addition, TLR4^−/− mice have exaggerated levels of activated caspase-3 expression in lungs, as assessed by Western blot analysis, during hyperoxia compared with WT mice (Fig. 3). These data indicated that the mechanism of increased lung apoptosis and injury in TLR4^−/− mice was the lack of key antiapoptotic responses, namely Bcl-2 and phospho-Akt induction, and increased proapoptotic processes such as caspase-3 activation. TLR4 signaling is generally thought to be proapoptotic, but this has previously been studied in the context of LPS, microbes, or immune cells (20, 21, 22). We show both in vivo and in epithelial and endothelial cells isolated from TLR4^−/− mice that TLR4 is necessary to maintain appropriate antiapoptotic responses during hyperoxia.

Hyperoxia likely induces lung TLR4 signaling via increased endogenous ligands such as fibronectin and hyaluronan in vivo. Alternatively, ROS may directly “ligate” TLR4 by affecting redox-sensitive moieties of the receptor given that hyperoxia can also induce TLR4 in isolated lung cell culture systems (data not shown). Others have demonstrated the involvement of the endothelial and epithelial cell in TLR signaling (23, 24). Potential explanations as to why TLR4 deficiency is protective in certain settings yet not in others include the use of different mouse strains and injury models. It is clear that distinct TLR signaling pathways are used in response to different types of injury. For instance, unlike LPS or ischemia-reperfusion-induced TLR signaling in which NF-κB mediates TLR signal transduction (25), NF-κB does not appear to have a major role in hyperoxia-induced injury (26).

HO-1 and its reaction products exert potent antioxidant and antiapoptotic properties in a variety of injury models (12, 27). We have demonstrated that HO-1 and its reaction product CO have the ability to induce the antiapoptotic proteins Bcl-2 and phospho-Akt (13, 28), both of which are lacking in the TLR4^−/− mice. Of note, TLR4^−/− mice have the ability to up-regulate HO-1 expression during hyperoxia. However, stress-induced HO-1 expression is likely a consequence of severe oxidant injury, and HO-1 levels as well as timing of induction in this context are inadequate to provide significant protection (12). High levels of HO-1 and its products are required before or at the onset of injury to have beneficial effects. Therefore, we administered intranasal rat HO-1 in an adenoviral vector (Ad-HO-1) to achieve HO-1 overexpression in mouse lungs before hyperoxia exposure. WT mice given Ad-HO-1 showed significantly increased survival during hyperoxia compared with WT mice given empty vector (Ad-LacZ) (Fig. 4,A). Furthermore, TLR4^−/− mice given Ad-HO-1 also showed markedly increased survival compared with TLR4^−/− mice given Ad-LacZ, reaching a survival rate comparable to WT/Ad-LacZ mice (Fig. 4 A). After 5 days of hyperoxia exposure, 50% of TLR4^−/− mice with Ad-HO-1 remained alive, whereas all of the TLR4^−/− mice with Ad-LacZ were dead. The range of survival for TLR4^−/− mice with Ad-HO-1 was 4–6 days, and for TLR4^−/− mice with Ad-LacZ the range of survival was 3.5–5 days. The improved survival with Ad-HO-1 correlated with attenuation of lung inflammation, permeability, and oxidative DNA damage in TLR4^−/− mice during hyperoxia (data not shown).

In addition, Ad-HO-1 rescued TLR4^−/− mice from hyperoxia-induced lung apoptosis (Fig. 4,B). As expected, Ad-HO-1 led to appropriately increased levels of lung HO-1 protein expression in TLR4^−/− compared with TLR4^−/− mice given Ad-LacZ before hyperoxia (Fig. 4,C). Interestingly, Ad-HO-1 restored levels of Bcl-2 and phospho-Akt protein expression in TLR4^−/− mice to that of WT mice when challenged with hyperoxia (Fig. 4 C). This indicated that TLR4^−/− mice retain the ability to induce cytoprotective pathways. Taken together, these data demonstrate that TLR4 is essential for survival, possibly by maintaining appropriate levels of antiapoptotic responses such as Bcl-2 and phospho-Akt induction in the face of oxidant stress. Although links between TLR4 and Akt have been described (19), the precise pathway(s) whereby TLR4 modulates Akt and Bcl-2 during oxidant injury is unknown and will be an important focus of future studies. There is also a possibility that TLR4^−/− mice are deficient in multiple signaling pathways, which thus opens up new avenues of investigation.

Our current data point to a novel paradigm for TLR4 signaling in response to exogenous oxidants. TLR4, at least in the lung, appear to be not only important sensors of conserved microbial components but also exogenous oxidants and thereby modulators of downstream responses such as inflammation and apoptosis. This would make teleological sense for an organ that is the first-line defense against inhaled microbial Ags as well as numerous oxidative stressors in the environment. In addition, the mechanisms of TLR4 responses in the lung during oxidant stress are likely to be distinct from what has thus far been described in other systems. Ultimately, these studies bring forth a fundamental link between innate immunity pathways and exogenous, noninfectious injury signals. These new insights will allow broader approaches for the prevention and treatment of a variety of oxidant-mediated disease processes.

We thank Suping Chen for her technical assistance, Juan Fan for mouse breeding, and Susan Ardito for her administrative assistance.

The authors have no financial conflict of interest.